Comparison of Classification Algorithm to Determine Potential Fishing Area Based on Coral Reefs

Morotai Island has a high potential for tuna, but until now, its utilization still needs to be considered optimal. Many factors cause this condition, in addition to the limited fishing facilities and infrastructure, the lack of proper fishing management, the lack of human resources, and the difficulty of detecting potential fishing locations, especially in certain seasons. This research aims to classify potential fishing locations on Morotai Island. The method used remote sensing using sentinel two and, in this research, used two algorithms as a comparison: linear discriminant analysis and random forest. The results of this study obtained that the LDA index value has a range of about 0.25 - 0.65, which means that the detection is not going well because this value means that the vegetation has a low chlorophyll content. While the random forest has a value range of 0.57 - 0.90, which means the chlorophyll detected on coral reefs has high chlorophyll, and it was obtained that the forest classification random obtained an accuracy of 95% and an LDA of 69%. This comparison shows that a random forest is better for testing the data classification.